Unverified50% confidenceFactExact time
GPT-4、Claude等通用大模型在预训练阶段学习的是互联网上的海量公开文本，倾向于给出通用的回答
1
Sources
50%
Confidence
Long-term
Relevance
7/16/2026
First Seen
Sources
Related Claims
UnverifiedThe training corpora of mainstream large language models such as the GPT series and Claude are primarily crawled from internet content dating from the 2000s onward.81% similarUnverifiedMany open-source models use high-quality outputs generated by commercial models such as GPT-4 and Claude as training data, which is the knowledge distillation training paradigm.80% similarUnverified主流模型如 GPT-4、Claude 在训练阶段被强化了对系统提示词的优先遵从性，系统提示词的约束效力通常强于用户指令79% similarUnverifiedMainstream large language models (including GPT-4, DeepSeek, Claude, etc.) are all trained on static corpora and have a definitive knowledge cutoff date.79% similarUnverified大语言模型如GPT-4、Claude通过在海量人类文本上进行预训练，习得了模拟人类对话风格、情感表达和个性特征的能力78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/531489API
curl https://kongchang.com/api/v1/knowledge/claims/531489MCP
get_claim(id=531489)